> Definitely a bug in Polipo (which doesn't mean it's not a bug in Firefox > too). > > CLOSE_WAIT means that one endpoint did a graceful close on a connection > but then other end didn't. Since Polipo is supposed to forcibly close > all idle connections after a timeout, a CLOSE_WAIT connection should not > remain forever.
I thought I had reported a similar problem before but cannot find it, anyways... When I fire off any number of simultaneous connections through polipo (say 50-250++) it often stacks up open file descriptors on network connections and never closes them down again. Even when whatever the the apps I spawned off (wget) have long since exited, polipo still holds out the list. This blocks further use when it happens. A kill and restart are needed. This occurs on the client side, not the server side. For instance, I've got a thousand stacked and locked connections in one sockstat output. I will try to instrument packet cap and netstat when I can. # socksProxyType=socks5 logLevel=0xFF # clientTimeout=190 serverIdleTimeout=180 serverTimeout=180 These don't seem to do or say anything useful about it. I do not see a client idle timeout setting, for which you indicate a possible presence and action... ------------------------------------------------------------------------------ Keep Your Developer Skills Current with LearnDevNow! The most comprehensive online learning library for Microsoft developers is just $99.99! Visual Studio, SharePoint, SQL - plus HTML5, CSS3, MVC3, Metro Style Apps, more. Free future releases when you subscribe now! http://p.sf.net/sfu/learndevnow-d2d _______________________________________________ Polipo-users mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/polipo-users
